Research On Nitrogen Nutrition Diagnosis Of Wheat Using Digital Image Processing Technique

Quick nitrogen diagnosis to crops can timely corresponding fertilization measures for diagnosis.Studies have shown that the leaf color can reflect crop nitrogen levels.In order to quickly guide wheat nitrogen fertilization at jointing stage,we use computer vision technology to diagnose wheat nitrogen nutrition.In this research,wheat canopy images of different cultivars and different nitrogen levels were taken by different cameras in different photograph programs.Wheat canopy image under the natural light has the feature of nonuniform illumination,complicated background with shadows and Classic image segmentation algorithm had low-accuracy and over-segmentation problems.In order to determin suitable characteristic parameters used on wheat nitrogen diagnosis,the wheat color characters were extracted and quantized to analyze the correlation between wheat leaf color characteristic parameters and the leaf nitrogen concentration(LNC)or chlorophyll concentration(or SPAD value).Most crop nitrogen nutrition diagnosis were based on a single shooting plan or without shoot plan,image acquisition had no unified standard.For the above problems,the contributions of this paper are as follows:(1)A k-means clustering algorithm based on HSI color space was proposed and it had had accurate segmentation results under different illuminations.The wiener filtering algorithm was used to deblur the blurred wheat canopy image caused by camera shake in the image preprocessing stage.And R+G-B was used to normalize color images in RGB space for the purpose of widening the gap between background and the target image.Wheat canopy image under the natural light has the feature of nonuniform illumination,complicated background with shadows.To reduce the impact,the HIS color space which is suitable for the condition of uneven illumination was adopted.A k-means clustering algorithm based on HSI color space was proposed to conquer the problem of low accuracy and over segmentation existing in Classic image segmentation algorithm.After transforming the normalized image from RGB to HSI color space,the different methods of K-means cluster used to the H weigh depends on whether the sunlight was uniform or not.The target image was gained after using mathematical morphology and noise-removal process.The experiments showed that compared with K-means cluster processing in Lab space and otsu algorithm based on H weigh,the method based on H weight could avoid over segmentation and had accurate segmentation results in different N fertilization,different illuminations and different periods.(2)CMI was proposed and determined to be a wheat nitrogen nutrition diagnosis parameter.By analysing the ralationship between nitrogen nutrition diagnosis parameters and all values of the CMI extacted from wheat canopy images,we got the approximate optimal combination of CMI = a*R + b^G + c*B.By analysing the correlation between CMI and nitrogen nutrition diagnosis parameters under different planting plans,this study determined the Normalized Red Index(NRI)value and the CMI value we proposed were suitable for the wheat nitrogen nutrition diagnosis.In the experiment,we choose two wheat cultivars,Yangmai NO.18 is more suitable for nitrogen nutrition diagnosis using digital image processing technology than Shengxuan No.6.The CMI parameter was more suitable for nitrogen nutrition diagnosis of Yangmai NO.18,CMI and NRI parameter were both suitable for nitrogen nutrition diagnosis of Shengxuan No.6.(3)By analysing the correlation between image charecteristic parameters and nitrogen nutrition diagnosis parameters,this paper determined the height of 1.0m and angle of 90 degree as the best shoting height and angle.This paper designed different photograph schemes to study the impact of correlation diagnosis.The schemes are including setting different photography heights,angles and different cameras.It proved that the correlation coefficient R2 was the best when the camera had the distance of 1 m above the wheat canopy and perpendicular to the ground.Also the SLR camera had a better nitrogen nutrition diagnosis effect than ordinary cameras.The above-mentioned research results could provide theoretical basis and technical support for the wheat canopy images under natural light of feild environment.lt determined the suitable photography height and angle and put forward a new guiding wheat nitrogen nutrition diagnosis parameters of CMI,and had important theoretical and practical value.